How to Test Code That AI Wrote

The AI Builds Fast. It Does Not Write Tests.

AI builders generate working apps quickly. But they skip tests. For a side project, that might be fine. For anything with real users, you need some level of testing to catch bugs before your users do.

You do not need 100% coverage. You need tests on the parts that would hurt most if they broke.

What to Test First

Authentication Flows

Login, signup, password reset, and session handling. If auth breaks, nobody can use your app. Test that:

• Users can sign up with valid credentials
• Invalid emails get rejected
• Password reset emails send correctly
• Logged-in users stay logged in across page refreshes

Payment Flows

If your app charges money, test the payment path end to end. A bug here costs you real revenue.

Core Business Logic

Whatever your app's main feature is, test that. A task manager should test creating, completing, and deleting tasks. A booking app should test making and canceling reservations.

Simple Testing Setup

Next.js apps work well with Vitest for unit tests and Playwright for end-to-end tests.

Install Vitest:

npm install -D vitest @testing-library/react

Write a basic test:

import { expect, test } from 'vitest'

test('formats price correctly', () => {
  expect(formatPrice(1999)).toBe('$19.99')
})

For end-to-end tests, Playwright runs a real browser:

npm install -D @playwright/test
npx playwright install

Ask the AI to Write Tests

Here is the shortcut: after the AI builds your app, ask it to write tests for the most important flows.

"Write Vitest tests for the authentication utilities." "Write a Playwright test that signs up, logs in, creates a task, and marks it complete."

The AI is good at generating tests when you tell it what to cover. Review the tests to make sure they actually check meaningful behavior, not just that the page renders.